I found many function generating prime numbers but all of them uses the knowledge of n-1 prime numbers that we already found or in a more inefficient way.
My question is not about existence of prime generating function but about the efficiency of such functions. Neglecting all other overheads, just the need to use all previously found primes making any algorithm to have $\Omega(n^2)$ running time.
Is there any efficient formula? If not, is there any argument supporting that you have to use all n-1 primes?